Why Your Network Should Go
IPv6-Only
Jeff Loughridge
Owner, Brooks Consulting
April 11, 2012
Overview
•
•
•
•
•
Early planning for IPv6 adoption
Dual stack complexity and cost
The argument for IPv6-only
IPv6-only challenges
How you can help IPv6 adoption by
promoting IPv6-only
2
The Best Laid Plans
• Early planning for IPv6 adoption assumed
– Gradual deployment
– Deployment complete before IPv4 exhaustion
– Operational experience would increase over a
series of years
• Dual stack was the ideal method for IPv6
deployment…in 2001
3
2012 – The Reality
• IPv4 addresses depleted or nearly depleted
• Immature IPv6 ecosystem (think CPE
routers)
• Limited operational experience with IPv6
• Doubled down on NAT to extend IPv4’s
lifetime
4
Dual Stack
Advantages
• Access IPv4 and IPv6
content without
mechanisms for interoperability
• Allows operators to
gather experience with
IPv6
Disadvantages
• Does not solve IPv4
address exhaustion
• Increased complexity
and cost
• Unexpected
interaction between
protocols
5
Complexity of Dual Stack
• Double efforts
– Configuration – ACLs, routing policy, QoS
– Training
– Management of two routing protocols (unless
you use IS-IS)
• Troubleshooting challenges
– Unforeseen interactions between IPv4/IPv6
– Protocol selection pushed to application layer
(e.g., Happy Eyeballs)
6
Costs of Dual Stack
• Staffing – workload increases, so do your
costs
• Network components- control cards in
wireless packet core, firewalls, service cards
• Software licensing
7
Example of Increased Cost for DS 3GPP Licensing
• Packet Data Protocol (PDP) context—
known as Evolved Packet System (EPS)
bearer in LTE—contains information about
the mobile session
• PDP contexts/EPS bearer types
– Pre-Release 8 – IPv4, IPv6 for GPRS
– Release 8 – IPv4, IPv6, IPv4v6 for EPS/LTE
– Release 9 – IPv4v6* for GPRS
* Not widely deployed
8
Costs to the Operator Increase
• Mobile network operators pay licensing
fees based on the number of activated PDP
contexts or PDN connections
• Dual stack increases number of PDP
contexts, thus increasing costs
9
10
Working Backwards from End State
End-to-end IPv6 +
NAT64/DNS64 for ~50%
of flows (Possible today)
IPv4-only or Dual
Stack
End-to-end IPv6
End-to-end IPv6 +
NAT64/DNS64 for
stragglers
11
Why IPv6-Only?
• Going IPv6-only puts you closer to the
Internet’s planned future state
• Avoids additional IPv4 address
consumption and business risk of being
tied to IPv4 address availability
• Avoids expensive intermediate steps,
especially in the data center
12
More Benefits of IPv6-Only
• Illuminates IPv6 bugs and feature gaps
otherwise hidden by dual stack
• Allows you to address these issues prior to
widespread deployment
• Drives down the cost of NAT as IPv6
content increases
• Enables growth of cloud, M2M, mobile
13
IPv6-Only Challenges
• No ideal method for IPv4-IPv6 translation
• IPv4 literals passed in application data or
binding directly to IPv4 addresses breaks
any translation
• Problematic apps – VoIP apps and gaming
Corner cases should not delay the progress
of IPv6-only. We can’t wait for an ideal
solution.
14
How You Can Help IPv6 Adoption
by Promoting IPv6-Only
• Gather production IPv6-only experience
and share with the community
• Drive the OS and application developers to
fix bugs and add needed features
• Ensure your services work for IPv6-only
hosts
15
Conclusion
• Dual stack is not the answer
• IPv6-only is our target end-state
• Deploying IPv6 end systems has
technology, risk avoidance, and cost
benefits
Challenge: Deploy IPv6-only end systems in at
least a segment of your network
16
Jeff Loughridge
jeffl@brooksconsulting-llc.com
http://brooksconsulting-llc.com
+1 703-229-0098
17